OBJECTIVES: Many epidemiological studies have analysed the relation between hormonal replacement therapy and risk of breast cancer. We present here a synthesis of the available results. MATERIAL AND METHODS: In 1997, a meta-analysis synthesised 90% of the data available at the time, obtained from 15 cohort and 36 case-control studies. This meta-analysis collected 53,000 breast cancers, 22% of these breast cancers having been observed in cohort studies. Since the publication of this meta-analysis, six studies totalling 8,000 cases (32% from cohort studies) have been published: two of these studies were partially included in the meta-analysis and the other four provided entirely new data. RESULTS: The conclusion, based on the available evidence, is that the risk of breast cancer diagnosis is higher among women who have used hormonal replacement therapy than among women who have not. The risk increases with treatment duration, is reduced when treatment is stopped and disappears almost completely a few years (5 in the meta-analysis) after the end of the treatment. The increase in the risk of breast cancer may be larger with estrogen-progestogen therapy than with estrogen alone. What are the consequences in terms of public health? From the results of the meta-analysis, one can estimate that the use of replacement therapy for 10 years between age 50 and 60 by 1,000 women will lead to the diagnosis of 6 extra breast cancers. This corresponds to the observation of 69 cases per 1,000 treated women after 20 years follow-up, as compared to the 63 cases expected among 1,000 untreated women. DISCUSSION: These results require confirmation. The observed increase in the risk of breast cancer is open to biases, particularly to a screening bias if women receiving hormonal replacement therapy have a more intensive surveillance than other women. Overall, the total number of women, across studies, who have received estrogen-progestogen therapy is small and their treatment was usually a combination of natural estrogens and medroxy progesterone acetate; therefore, the long term consequences of the treatments commonly used in France (synthetic estrogens and a variety of progestogens) have not been evaluated, this evaluation is much needed. CONCLUSION: What is needed is a global view of the risks and benefits of hormonal replacement therapy, and more epidemiologic data are required to reach this goal. In the present state of knowledge, the moderate excess risk of breast cancer observed does not justify a change in medical practice, neither in the selection of women to whom hormonal replacement therapy should be offered nor in the duration of their treatment.